Setup

Packages

Data Import

Normalisation

Samples

New samples:

59 samples in total

Including:

Old and young male and female pools

Old male and female individuals

Females at 0, 10 and 20 days after moulting, pools and individuals

Regeneration experiment sets: each pool individual has three samples taken:

Left T4 & T5 limbs at t0 (bef)

Left T4 & T5 limbs at t30, after regeneration (reg)

Right T4 &T5 limbs at t30 (un)

Sample Quality

Quality is somewhat variable, with some samples, especially those with small amounts of RNA giving poor results.

Is the age signal retained?

Females only

Can we replicate the previous result? (Two low quality samples removed)

Yes! Our new samples cluster correctly with their age groups.

And if we add the males and repeat the PCA:

No good! The same set of markers doesn’t function to separate old and young in males as it does in females.

Males only

Let’s try to find an aging signature amongst the males. Using successive log2 old vs young foldchange cut-offs doesn’t work, but we can try using our successive t-test p-value cut-off approach.

So we also have an aging signal for the males, looks best @ pval<0.1 (10 markers). How does this compare to that for the females? Are these a subset of the markers that show a signal amongst females?

Male vs Female Aging Signal

So males don’t show the same aging signature as seen amongst the females. Why is this?

Age signal by marker

Females

Let’s look at our female aging signal in more detail.

We can also examine this aging signal on a marker by marker basis. Using our 61 chosen markers:

In our complete marker set:

A strong age signal is present, but there is still quite a lot of noise present.

Males on a per marker basis

So we do have an aging signal in males, albeit a weaker one, present in fewer markers.

Variation in old vs young

In the above graphs, the old seem more variable, lets test this. Is it just a perception due to the log scale on the graphs above?

Yes!

Effect of Regeneration

Regen effect on PCA

Do we see a rejuvenation effect? Here we take our female old and young PCA, then project the regenerated samples (Nikos and mine) onto it.

Let’s compare the regenerated samples with their controls

We don’t see an overall rejuvenating effect. Maybe examine this on a marker by marker basis?

Regen effect on per marker basis

The regenerated samples look to be much more variable. How does it look on the density plot?

The wide range and “shoulders” of the regenerated samples in the green may provide support for my previous idea of the “hyper young” and “hyper old”

Regeneration experiment in detail

This might be clearer if we are comparing the regenerated samples with their controls. Only our regen sets: three samples (bef, un, reg) for each pool/individual ##PCA of regeneration sets only

No clear pattern visible here.

Per marker effect

Lets compare the unamputated with the regenerated samples.

Marker expression is all over the place after regeneration, not sure what to make of this. Could be that this is the “hyper young and”hyper old" marker response to regeneration?

Could there be a better way of normalising this data? This could help to cut through the noise and reveal our signal.

Compare to the regen effect from the previous dataset

Let’s add the values for Nikos’ regenerated samples (normalised to their unregenerated partners) to this plot and see if they fit with what we find for our new samples.

The effect in Nikos’ dataset doesn’t seem to fit very well with what we have here. The important issue to try to get at here, I think is the degree to which each marker’s response is stereotyped. Do any makrers respond reliably and stereotypically to regeneration? How best to examine this?

Moulting effect

What effect does the moulting cycle have?

Seemingly, recently moulted =younger. Response is variable though!